HAL Id: inserm-03169103
https://www.hal.inserm.fr/inserm-03169103
Submitted on 15 Mar 2021
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
melanogaster
Sophie Layalle, Laetitia They, Sarah Ourghani, Cédric Raoul, Laurent
Soustelle
To cite this version:
Sophie Layalle, Laetitia They, Sarah Ourghani, Cédric Raoul, Laurent Soustelle. Amyotrophic Lateral
Sclerosis Genes in Drosophila melanogaster. International Journal of Molecular Sciences, MDPI, 2021,
22 (2), pp.904. �10.3390/ijms22020904�. �inserm-03169103�
Review
Amyotrophic Lateral Sclerosis Genes in Drosophila
melanogaster
Sophie Layalle1, Laetitia They1, Sarah Ourghani1, Cédric Raoul1,2,* and Laurent Soustelle1,*
Citation: Layalle, S.; They, L.; Ourghani, S.; Raoul, C.; Soustelle, L. Amyotrophic Lateral Sclerosis Genes in Drosophila melanogaster. Int. J. Mol. Sci. 2021, 22, 904. https://doi.org/ 10.3390/ijms22020904
Received: 21 December 2020 Accepted: 14 January 2021 Published: 18 January 2021
Publisher’s Note:MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil-iations.
Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).
1 The Neuroscience Institute of Montpellier, INSERM, University of Montpellier, 34091 Montpellier, France; sophie.layalle@inserm.fr (S.L.); laetitia.they@inserm.fr (L.T.); sarah.ourghani@inserm.fr (S.O.)
2 Laboratory of Neurobiology, Kazan Federal University, 420008 Kazan, Russia
* Correspondence: cedric.raoul@inserm.fr (C.R.); laurent.soustelle@inserm.fr (L.S.)
Abstract:Amyotrophic lateral sclerosis (ALS) is a devastating adult-onset neurodegenerative disease characterized by the progressive degeneration of upper and lower motoneurons. Most ALS cases are sporadic but approximately 10% of ALS cases are due to inherited mutations in identified genes. ALS-causing mutations were identified in over 30 genes with superoxide dismutase-1 (SOD1), chromosome 9 open reading frame 72 (C9orf72), fused in sarcoma (FUS), and TAR DNA-binding protein (TARDBP, encoding TDP-43) being the most frequent. In the last few decades, Drosophila melanogaster emerged as a versatile model for studying neurodegenerative diseases, including ALS. In this review, we describe the different Drosophila ALS models that have been successfully used to decipher the cellular and molecular pathways associated with SOD1, C9orf72, FUS, and TDP-43. The study of the known fruit fly orthologs of these ALS-related genes yielded significant insights into cellular mechanisms and physiological functions. Moreover, genetic screening in tissue-specific gain-of-function mutants that mimic ALS-associated phenotypes identified disease-modifying genes. Here, we propose a comprehensive review on the Drosophila research focused on four ALS-linked genes that has revealed novel pathogenic mechanisms and identified potential therapeutic targets for future therapy. Keywords:amyotrophic lateral sclerosis; Drosophila melanogaster; SOD1; C9orf72; FUS; TDP-43
1. Introduction
ALS (amyotrophic lateral sclerosis) also known as Charcot’s disease or Lou Gehrig’s disease is a fatal adult-onset neurodegenerative disease affecting the motor system [1–4]. ALS is the most common motoneuron disorder with an incidence of two per 100,000 individuals, which varies according to geographical differences, and a mean onset at 65 [5,6]. ALS is characterized by the progressive dismantling of the neuromuscular junctions and degeneration of motoneurons in the brain and spinal cord [7,8]. Motoneuron loss leads to progressive paralysis and death due to respiratory failure within 3 to 5 years of onset disease [9]. In addition to motoneuron degeneration, ALS is clearly a non-cell-autonomous disease as astrocytes, oligodendrocytes, microglial cells, and blood-derived immune cells also contribute to the selective degeneration of motoneurons [10]. Moreover, ALS forms a broad neurodegenerative disease continuum with frontotemporal dementia (FTD) disease, and up to 50% of ALS patients concomitantly develop cognitive impairment or behavioral changes [11–13]. Despite recent promising gene therapy approaches, no effective cure is currently available for ALS patients [14]. Most cases of ALS are sporadic (sALS), and up to 10% have been classified as familial ALS (fALS) [15]. Currently, fALS-associated mutations have been found in approximately 50 genes, and more than 30 are thought to be causatives [16,17]. The most commonly mutated ALS-linked genes are superoxide dismutase-1 (SOD1), chromosome 9 open reading frame 72 (C9orf72), fused in sarcoma (FUS), and TAR DNA-binding protein (TARDBP) [18].
The antioxidant enzyme SOD1 was the first gene linked to fALS in 1993 [19]. This gene encodes a Cu/Zn superoxide dismutase, whose function is to catalyze the conversion
of the superoxide ion, a toxic reactive oxygen species (ROS) produced during cellular respiration, to dioxides [20]. In 2011, abnormal GGGGCC hexanucleotide repeat expansion (HRE) within the C9orf72 gene was identified as a new cause of ALS and frontotemporal dementia (FTD) [21,22]. Currently, intronic HRE in the C9orf72 gene represents the most common genetic cause of ALS [23]. C9orf72 is part of a guanine nucleotide exchange factor complex [24], whose precise function remains unclear, but which was shown to be an important regulator of membrane trafficking and autophagy [25]. Lastly, FUS and TARDBP encode two DNA/RNA-binding proteins, which play distinct roles in transcription, as well as numerous roles in RNA metabolism, including splicing, stability, and transport [26]. FUS is a protein belonging to the heterogenous nuclear ribonucleoproteins (hnRNPs) (also known as hnRNP P2) [27], belonging to the FET protein family that includes two other RNA-binding proteins (RBPs) EWS and TAF15 [28]. In 2009, FUS was identified to be involved in fALS cases [29,30]. TARDBP encodes the TDP-43 protein, which is mainly nuclear and shuttles between the nucleus and the cytoplasm. Nuclear depletion and cytoplasmic aggregation of TDP-43 are found in most if not all ALS patients independently of the mutated status of TDP-43, making it a hallmark of the disease [31]. However, it is still debated whether TDP-43 cytoplasmic aggregation is deleterious or protective for ALS disease [32].
Drosophila melanogaster is a model easy to handle, cost-effective, with a short lifespan and a fully sequenced genome since 2000 [33,34]. In addition, Drosophila is a power-ful genetic model with several genetic tools, such as the upstream activating sequence (UAS)/Gal4 system (Figure1) [35], which is extensively used to overexpress Drosophila or disease-associated human genes in a tissue/cell-specific manner. Combined with the temporal and regional gene expression targeting (TARGET) or gene-switch systems [36] (Figure1), gene expression can be controlled temporally allowing to investigate behavioral studies, avoiding developmental alterations. Furthermore, pan-genomic screenings, using RNA interference (RNAi)-induced gene knockdown for example, have been successfully used to identify genetic modifiers of human disease-associated phenotypes. It is estimated that as many as 77% of the human disease-associated genes have fly orthologs [37]. Fur-thermore, 76% of human proteins involved in synaptic vesicle trafficking have a Drosophila ortholog [38], indicating that synaptic transmission machinery is well conserved in flies. For all these reasons, Drosophila has emerged as a powerful genetic model for studying several neurodegenerative diseases (for reviews, see [39–41]) (Figure2). Genetic studies in Drosophila have provided novel insights into the cellular and molecular mechanisms of ALS-linked neurodegeneration. Here, we review the Drosophila models that have been developed to better understand the function and decipher the pathological consequences associated with SOD1, C9orf72, FUS, and TARDBP genes.
Figure 1. Drosophilagenetic tools. The Gal4 system introduced in Drosophila allows transgene ectopic expression with spatiotemporal control. The upstream activating sequence (UAS)/Gal4 is a bipartite system. One transgenic line expresses the Gal4 transcription factor from yeast in a tissue-specific manner using an endogenous promoter or an enhancer sequence. The other line carries the transgene under the control of the UAS (upstream activating sequence). In the progeny of the cross, the UAS is bound by Gal4 protein and the transcription of the gene of interest starts. The temporal and regional gene expression targeting (TARGET) method allows a more flexible temporal control of the Gal4 system. A temperature-sensitive version of Gal80 protein (Gal80TS) is expressed ubiquitously
under the control of the Tubulin promoter. At permissive temperature (18◦C), Gal80TS bound
to Gal4 and prevents the starting of the transcription. A temperature shift to 29◦C relieves the
transcriptional repression and permits the transcription of the gene of interest. The GeneSwitch system is based on hormone-inducible Gal4 and allows a temporal regulation of transgene expression. Gal4 is fused to a progesterone receptor, and the addition of hormone or synthetic ligand (RU486) in the feeding of the fly (adults or larvae) activates the transcription of the gene of interest. This system is reversible as the removal of ligand silent the system. All these genetic tools are routinely used in laboratory to precisely control transgenes expression not only in a specific tissue but also in a precise temporal manner.
Figure 2. Drosophilais a model to study neurodegenerative diseases. Several neurode-generative diseases, including amyotrophic lateral sclerosis (ALS), have been modeled in Drosophila via transgenic expression of wildtype or mutated human proteins. The toolkit available at Drosophila allows an in-depth study of the neurodegenerative mechanisms associated with these diseases. Several behavioral tasks, such as larval crawling and climbing assays, allow monitoring the locomotor activity during Drosophila life. Drosophila lifespan assays are useful to follow the time course of neurodegeneration and might be used as a readout for genetic screens. When expressed in the eye, toxic proteins disrupt the stereotyped organization of ommatidia and bristles, leading to a rough eye phenotype. This easily observable readout allows to perform genetic screens aimed at identifying modifiers (enhancers or suppressors) of the rough eye phenotype. Dendritic arborization neurons have a complex dendritic branching pattern and are widely used to identify genes involved in dendrite morphogenesis processes. The larval neuromuscular junctions (NMJs) are glutamatergic synapses that use ionotropic glutamate receptors and post-synaptic scaffolding proteins sharing similarities with mammalian brain synapses. These NMJs are easy to visualize and relatively simple with few axonal branches composed of synaptic boutons, which contain the active zones (sites of neurotransmitter release). Mor-phological and electrophysiological analyses on larval NMJs are frequently used to study how gene loss or gain of function might influence synapse development and function.
2. SOD1 Gene
2.1. dSod1 and the Aging Theory
In the 1980s, the concept of the aging process emerged with the assumption that aging reflects over time an accumulation of changes associated with an increasing susceptibility to develop diseases and, finally, death [42]. The free-radical theory assumed that the basic cause of aging lies on the deleterious effects produced by free-radical reactions [43–46]. This theory was supported by the observation that caloric restriction or lowering the metabolic rate decreases free-radical production, increasing the average lifespan in different species including Drosophila [47]. The prediction was to extend lifespan by increasing antioxidant
levels to overcome the damages caused by free-radical reactions. As the main source of free radicals in aerobic eukaryotes is generated by oxygen metabolism, in that context, many studies have focused on the dSod1 gene. At that time, the link between ALS disease and the mutations in hSOD1 was not known, and the idea was to genetically increase lifespan in animal models especially in Drosophila using dSod1.
In Drosophila, the gene dSod1 was cloned in 1989 [48]. Many studies have shown that different genetic conditions leading to dSod1-null mutants, such as deletions or missense mutations inactivating the enzymatic activity of the protein, exhibited several phenotypes. The lifespan was drastically reduced by 85–90% and the locomotor activity was also im-paired. The resistance to oxidative stress conditions was lowered in dSod1-null animals, whereas abnormal wing morphology and infertility were also observed [49–51]. The dSod1 loss-of-function (LOF) phenotypes were unsurprising for a ubiquitous housekeeping en-zyme involved in detoxification and remained in the line with predictions and aging model. More unexpected were the consequences of dSod1 gain of function (GOF). Transgenic Drosophila lines carrying an additional copy of dSod1 were constructed using random P element insertion [52]. These flies showed 30–40% increase in the dismutase activity but with a minor effect on both lifespan and oxidative stress resistance (Table1). One expla-nation was that the increase in dSod1 activity level was too low to markedly increase the maximal lifespan of these flies. On the other hand, a previous study using transgenic flies expressing the bovine form of SOD1 under the control of actin5C promoter showed that ubiquitously increasing bovine SOD1 expression to high levels led to a deleterious effect with flies that did not emerge from their pupal cases [53].
Table 1.Phenotypes observed in dSod1 (superoxide dismutase-1) mutant backgrounds.
Mutant Line Phenotype Reference
dSod1n108
Homozygous lethality with rare eclosing adults, sterile, and early dying within 2–3 days No detectable superoxide dismutase activity
Hypersensitivity to paraquat
[49]
Necrotic lesions throughout retina [50]
dSod1x16 No detectable superoxide dismutase activityPartially lethal [[5051]]
dSod1x139
No detectable superoxide dismutase activity [50]
Partially lethal [51]
Reduced eclosion rate, eclosing adults with shorter lifespan [54,55]
dSod1x16/dSod1x139 Very reduced lifespan
Impaired adult locomotion [56]
dSod1G37R No lethality, adult eclosion as wildtype, no adult locomotion defect. [57]
dSod1G51S Impaired larval crawling
Reduced adult viability, adult escapers with locomotor defect [57]
dSod1G85R
Impaired larval crawling
Reduced adult viability, mostly lethal at pharate stage Adult escapers with muscle atrophy and denervation
[57]
Reduction in NMJ bouton number and electrophysiological defect [58]
dSod1H48R Reduced adult viability, mostly lethal at pharate stage [57]
dSod1H71Y
Impaired larval crawling
Reduced adult viability, mostly lethal at pharate stage, adult escapers with locomotor defect, muscle atrophy, and denervation
[57]
This table describes the phenotypes associated with different dSod1 mutant lines. Note that the experiments could be done at different temperatures.
2.2. Drosophila as a Modeling Tool to Understand hSOD1-Induced ALS 2.2.1. hSOD1 Was the First Gene Linked to ALS Disease
In 1993, the discovery of the genetic linkage between hSOD1 and the fALS [19] changed not only the vision of the disease but also the way the pathogenicity of SOD1 gene was considered. Indeed, to date, about 200 ALS-associated mutations in hSOD1 have been described [59]; most of them are missense point mutations (Figure3). The first lines of research considered to explain the death of motoneurons characteristic in ALS were oriented toward an LOF hypothesis. The mutated form of hSOD1 protein could lead to a decrease in the superoxide dismutase enzymatic activity, implying an impairment of free-radical elimination, leading to an LOF phenotype. Nevertheless, hSOD1 mutants showed different degrees of alteration in superoxide dismutase activity. For example, hSOD1G93A (at position 93, a glycine is substituted to an alanine) showed an identical enzymatic activity compared to wildtype protein [60]. In mice models expressing the G93A or G37R mutation in hSOD1, the level of dismutase activity was not altered but motoneuron degeneration still took place [8,61]. In addition, the lack of phenotype observed in mSOD1 knockout mice [62] strengthened the idea that motoneuron death in ALS is not only from a loss in dismutase activity. Indeed, it was shown that the mutations found in hSOD1 induce the misfolding of the protein and confer new noxious properties. This toxic GOF is worsened by the ability of misfolded hSOD1 to spread from cell to cell causing the propagation of the disease [63–73].
2.2.2. Gain-of-Function Drosophila Models
As we mentioned above, dSod1-null mutants showed a severe reduction in lifespan. In a dSod1-null background (dSod1x16/sodx39), expression of hSOD1WTunder the control of endogenous cis-regulatory dSod1 sequences fully rescued the lifespan reduction as dSod1 expression did. In this genetic background, the expression of different fALS-related hSOD1 mutant alleles (hSOD1A4V, hSOD1G37R, hSOD1G93C, hSOD1G41D, hSOD1I113T) showed a partial rescue of the lifespan compared to dSod1x16/sodx39flies. However, the lifespan of fALS-related hSOD1 flies was shortened compared to flies expressing hSOD1WTallele. This lifespan reduction was coupled with an early drop of negative geotaxis performance in line with pathological phenotypes [56]. This first study did not allow the targeting of hSOD1 expression specifically in motoneurons, and the level of expression was limited to the endogenous level of Drosophila Sod1. Drosophila models were, therefore, generated using the UAS/Gal4 system [35] to overexpress different hSOD1 transgenes (Figure1). Targeting wildtype hSOD1 or different mutated forms found in patients directly in motoneurons allowed analyzing the consequences of this expression at the whole animal level. Using a motoneuron Gal4 driver (D42-Gal4) to express either wildtype or ALS-related forms of hSOD1 (A4V or G85R), Nancy Bonini’s team showed that these different forms did not alter Drosophila lifespan [74] (Table2).
Figure 3.Structure of chromosome 9 open reading frame 72 (C9orf72) transcripts and SOD1, fused in sarcoma (FUS), and TAR DNA-binding protein (TDP)-43 proteins. The human SOD1 protein is composed of eight beta-strands (in purple) and connecting loops. The Cu-binding region, the Zn-binding loop, and the electrostatic loop are represented in green, yellow, and blue, respectively. ALS-associated missense mutations are located throughout the protein and are listed vertically
in pink dots. The human C9orf72 gene produces three variants. The (G4C2) hexanucleotide repeat expansion (HRE) is
located in the first intron of variants 1 and 3, and in the promoter region of variant 2. The human FUS protein constitutes a N-terminal glutamine/glycine/serine/tyrosine-rich region (QGSY-rich in dark yellow), a prion-like domain (PrLD in brown), a nuclear export signal (NES in blue), a RNA recognition motif (RRM in red), two arginine/glycine-rich regions (RGG1 and RGG2 in green), a zinc finger domain (ZnF in orange), and a C-terminal non classical nuclear localization signal (NLS in cyan). Most ALS-related mutations are found in the C-terminal NLS region of FUS protein. The human TDP-43 protein carries three mitochondrial localization domains (M1, M3, and M5 in pink), a nuclear localization signal (NLS), two RNA recognition motifs (RRM1 and RRM2 in red), a nuclear export signal (NES), and a glycine-rich domain (G-rich in brown). The ALS-associated mutations described in the review are indicated.
Table 2.Gain-of-function phenotypes induced by hSOD1. WT, wildtype.
Gal4 Line UAS Line Phenotype/Reference
D42-Gal4 (motoneurons)
hSOD1 WT No effect on lifespan, progressive motor dysfunction [74,75]
Abnormal synaptic transmission [74]
hSOD1 A4V No effect on lifespan, progressive motor dysfunction [74]
hSOD1 G85R
No effect on lifespan, progressive motor dysfunction [74,75]
Abnormal synaptic transmission [74]
Induction of stress response in glial cells [74]
D42-Gal4 (motoneurons)
3 mM BMAA
(β-N-methylamino-L-alanine)
hSOD1 WT Increased lifespan compared to control flies [76]
hSOD1 A4V Increased lifespan compared to control flies [76]
hSOD1 G85R Increased lifespan compared to control flies [76]
M1B-Gal4 (glial cells) 3 mM BMAA
(β-N-methylamino-L-alanine)
hSOD1 WT Accelerated death compared to control flies [76]
hSOD1 A4V Accelerated death compared to control flies [76]
hSOD1 G85R Accelerated death compared to control flies [76]
24B-Gal4 (muscles)
hSOD1 WT No phenotype in thoracic muscle fibers,
Slight motor behavior defect and normal lifespan [77]
hSOD1 G93A Upheld wings, swollen mitochondria,
Impaired motor behavior and reduced lifespan [77]
This table describes the phenotypes associated with different UAS lines. Note that the experiments could be done at different temperatures.
These flies displayed progressive motor function deterioration over time. Contrary to what is observed in vertebrates, no neuronal cell death has been demonstrated in the adult ventral nerve cord where the motoneuron cell bodies are located. Therefore, ALS Drosophila and mouse models showed different phenotypes and specificity. In mice, directing mutant hSOD1 to all neurons did not cause motoneuron disease [78–80], while motoneuron death is contingent on the ubiquitous expression of mutant hSOD1. In the same manner, the presence of mutant hSOD1 aggregates was not obvious in Drosophila nervous system compared to mouse models of ALS [68,70,81,82]. The insoluble hSOD1 inclusions cannot explain the observed motoneuron dysfunction. However, misfolded hSOD1 mutant proteins could be detected in flies using conformation-specific antibodies developed in mice [83,84]. Indeed, a study using the hSOD1G93Atransgene expressed under the control of the muscle specific driver (24B-Gal4) showed that misfolded hSOD1 protein was produced in muscles. This expression led to shortened lifespan, drop in motor activity, and mitochondrial impairment [77] (Table2). Unfortunately, this study did not investigate the effects of the hSOD1G93Atransgene expression in the nervous system. Interestingly, when hSOD1 mutant forms were cell-autonomously expressed by motoneurons, glial cells exhibited a stress response as evidenced by the expression of the heat-shock protein 70 (HSP70) [74]. The role of heat-shock protein in endoplasmic reticulum stress and, more particularly, the activation of the unfolded protein response (UPR) have been extensively studied in ALS disorder [85–88] (for a review, see [89]). Consistent with the important cross-talk between neurons and glial cells, recent studies suggested the idea of a non-cell-autonomous role of the UPR to modulate ALS progression [90–92]. The stress response induced in glial cells could, thus, help to provide support to motoneurons in the first stages of the disease. For example, HSP70 upregulation was protective during the disease progression in mouse models of ALS helping to maintain motoneuron innervation [93]. In a more general way, although the motoneurons are the main cells affected, ALS-related hSOD1 mutants in a non-cell-autonomous manner and the glial cells also play a part in ALS pathogenesis. The importance of glial cells was highlighted in BMAA
(β-N-methylamino-L-alanine) resistance study. BMAA is a neurotoxin found in cycad seeds that causes Guam
disease, an ALS–Parkinsonism dementia complex [94]. The expression of hSOD1 mutant (A4V, G85R) transgenes in either Drosophila motoneurons or glial cells led to different results. Expressed selectively in glial cells, the sensibility to BMAA was increased with a loss of motor performance over time. When hSOD1 constructs were expressed in motoneurons
or in both motoneurons and glial cells, the resistance to the neurotoxin was enhanced, as well as the motor functions [76]. The mechanism leading to these different phenotypes according to the targeted cell type remains unknown, but it shows that glial cells are involved in the course of the disease as documented in rodent models of ALS [63,95,96]. 2.2.3. Knock-In hSOD1 in Drosophila Led to Unexpected Results
The overall literature on the study of hSOD1 ALS-associated forms has shown that the phenotypes observed were not only dependent of the transgene expression level but were also related to the cellular type targeted. To overcome these difficulties and increase the understanding of the mechanisms leading to the toxicity of the SOD protein in ALS, recent studies chose to express hSOD1 at an endogenous level. For this, they used the ends-out homologous recombination strategy to replace wildtype dSod1 and introduce ALS-related mutations at conserved residues in dSod1, thereby creating dSod1G85R, dSod1H71Y, and dSod1H48R mutants [57]. In homozygous conditions, these mutants die throughout development, with escaper adult flies showing shortened lifespan and severe locomotion defects. Although these phenotypic traits are characteristic of ALS, the different mutants showed developmental defects, as dSod1G85Rhomozygous died at the pharate stage and did not emerge properly. Surprisingly, despite the severe locomotor defects observed, no motoneuron death was detected in larvae or adult flies. Interestingly, all mutants pro-duced significant amounts of dSod1 protein, and expression of wildtype dSod1 partially saved the dSod1G85R/G85Rphenotypes, leading to the hypothesis that both toxic GOF and LOF are combined to explain the observed ALS-associated phenotypes. Further study of dSod1G85R mutants at the neuromuscular junction (NMJ) level showed that both the number of boutons and the neurotransmission were reduced in the mutant, at the pharate stage [58]. Earlier in development, larvae dSod1G85Rdisplayed a clear crawling impairment that was not associated with a disturbing phenotype at the NMJ level. Electrophysiolog-ical recordings of NMJs and motoneurons did not show any difference in post-synaptic excitatory events between dSod1G85Rand wildtype flies, although motoneurons exhibited a slightly reduced excitability. The crawling behavior is generated by the central pattern generator network, an innate neural circuit composed of interneurons and motoneurons that generate rhythmic motor output. Sensory inputs play a role in adapting locomotor activity to external cues. Extracellular recordings and deafferentiation experiments showed that defective sensory feedback lead to reduced locomotor activity in dSod1G85Rlarvae [58]
(Table1). These findings highlight previous existing hypotheses on non-cell-autonomous factors involved in ALS degeneration [95].
2.3. ALS Drosophila Model to Test Neuroprotective Drug Candidates
Drosophila studies also allowed, in the ALS context with ectopic hSOD1 expression, to easily test different compounds for their protective or negative actions, on lifespan or motor activity for example. Thus, α-lipoic acid (LA) produced from plants and known for its various properties including antioxidant potential [97] has been tested in Drosophila expressing hSOD1G85Rin motoneurons. LA was shown to moderate the neurotoxicity,
extending lifespan and improving motor activity, in hSOD1G85Rflies [98]. In the same technical manner, γ-oryzanol (Orz), a component of rice bran oil known for its antioxidative activity was also tested [99]. In the Drosophila ALS context, Orz increased HSP70 expression and alleviated oxidative damage [75]. These results open the door to other studies to investigate the role of new drugs potentially neuroprotective for ALS disease.
Until now, studies concerning hSOD1 have focused on major phenotypes (longevity, motor deficit), but a more detailed analysis at the cellular level of the hSOD1 actions could provide answers on the pathways leading to the pathology. Moreover, future studies using Drosophila could help to understand the mode of propagation of misfolded hSOD1, as well as the relationships between different cell types (motoneurons, glial cells, muscles).
3. C9orf72 Repeat Expansions
In 2011, intronic GGGGCC (G4C2) HRE in the C9orf72 gene was identified as a new
cause of ALS [21,22]. In healthy individuals, the number of repeats is below 30, while it may reach thousands in ALS patients [100]. Actually, it is clearly established that abnormal HRE is the most common genetic cause of ALS [23], which accounts for around 40–50% of fALS and 5–10% of sporadic cases [23]. The human C9orf72 gene produces three alternative spliced transcripts [21] and, depending on the splice variant, the (G4C2) repeat is present
in the promoter region (for transcript variant 2) or within the first intron (for transcript variants 1 and 3) (Figure3). To date, the underlying mechanisms of neurodegeneration associated with C9orf72 HRE have been classified into three nonexclusive categories: (1) loss of C9orf72 function through haploinsufficiency, (2) sequestration of proteins by HRE-induced RNA foci, and (3) toxic GOF HRE-induced by dipeptide repeat proteins.
3.1. Loss of C9orf72 Function
Several studies have shown that the HRE in C9orf72 gene decreases the levels of messenger RNA (mRNA) [21,101] and protein [102,103] in patient tissues, suggesting that reduced C9orf72 protein function may play a role in the disease. C9orf72 LOF in Zebrafish [104] and Caenorhabditis.elegans [105] induced motor deficits, but degeneration of motoneurons was not observed in mice lacking C9orf72 gene [106]. However, a recent study showed that reduced C9orf72 function may exacerbate the repeat-dependent gain of toxicity [107], suggesting that the loss of C9orf72 function may be involved in the disease. The Drosophila genome does not contain a C9orf72 ortholog gene [108], making it impossible to determine the consequence of C9orf72 LOF in this organism.
3.2. Sequestration of Proteins by Expanded Repeat in RNA Foci
A hallmark found in different tissues, including motoneurons, of C9orf72-linked ALS patients is the presence of RNA foci [109]. These RNA foci are the result of bidirectional transcription of the HRE, leading to the accumulation of repeat-containing RNA aggre-gates, generally localized in the nucleus but also found in the cytoplasm. In Drosophila, it was found that components of the Drosophila DRB sensitivity-inducing factor (dDSIF) and Drosophila polymerase-associated factor 1 (dPAF1) complexes, two regulators of RNA polymerase II, are selectively required for the transcription of the expanded (G4C2) re-peats [110,111]. Indeed, RNAi-induced silencing of these components decreased the RNA production only from long expanded repeats, leading to reduced toxicity. One possible toxic effect of these RNA foci is that they may sequester RBPs, interfering with their func-tion and altering general RNA metabolism. For example, Xu et al. incubated biotinylated (G4C2)10 repeat RNA with mouse spinal cord lysates and identified Purα as a protein
able to specifically bind to the repeat [112]. Purα is an evolutionarily conserved RBP that modulates transcription and translation, and it is also a component of ribonucleoprotein granules [113]. The authors used a UAS construct carrying a (G4C2)30repeat cloned
up-stream of the translation start of GFP to express it in the developing eye. This induced a rough eye phenotype, revealing the toxicity of the expanded repeat that was not observed when the same construct carrying only 3 repeats was used. Overexpression of Purα with the (G4C2)30-GFP construct suppressed the neurodegeneration induced by the (G4C2)30-GFP
transgene, supporting the idea that Purα function was attenuated in the presence of the expanded (G4C2) repeat. When the (G4C2)30-GFP construct was expressed in motoneurons
using the OK371-Gal4 driver, around 50% adult eclosion failure and locomotion defect in 28 day old flies were observed. In addition, larval NMJs showed a reduction in active zone number. These phenotypes were suppressed by the overexpression of Zfp106, a zinc finger protein that binds specifically to a (G4C2)8repeat construct [114]. Interestingly, Zfp106
interacts with other RBPs including TDP-43 and FUS, suggesting that its sequestration by the expanded repeat may alter the function of ALS-related RBPs.
It was shown that G4C2HRE RNA forms hairpin and G-quadruplex structures that
(Ras-related nuclear GTPase-activating protein) that binds preferentially the sense RNA quadru-plex of the HRE [115]. RanGAP stimulates the hydrolysis of GTP to GDP carried out by Ran GTPase, which is involved in nucleocytoplasmic transport. The overexpression of RanGAP suppressed the neurodegeneration induced by (G4C2)30-GFP expression in the
de-veloping eye. This was also the case when eye-expressing (G4C2)30-GFP flies were fed with
5,10,15,20-tetrakis-(N-methyl-4-pyridyl)porphine (TMPyP4), a porphyrin compound that destabilizes RNA G-quadruplex tertiary structures. This indicates that the G-quadruplex structure was required for the recruitment of RanGAP on (G4C2) HRE RNA [115].
Burguette et al. found that (G4C2) expanded repeat RNAs, which generally aggregate
as nuclear RNA foci, also localize to neuritic granules that were actively transported along neuronal processes [116]. The presence of such hexanucleotidic repeat RNAs containing granules in neurites was associated with defects in neuronal branching, suggesting that they may confer local toxicity [116]. The class IV epidermal sensory dendritic arboriza-tion neurons of Drosophila have a complex dendritic branching pattern that was altered by the expression of a (G4C2)48 RNA repeat. The overexpression of dFMR1 or Orb2,
two transport granule components that regulate local translation, enhanced the dendrite branching defects, while their RNAi-induced knockdown rescued the phenotype [116]. This suggests that local translation may be affected by the presence of (G4C2) HRE RNAs
in neuronal processes.
Despite the putative sequestration of RBPs by HRE RNAs, several studies have demonstrated that RNA foci per se were not toxic. Tran et al. used a C9orf72 minigene containing (G4C2)160repeat flanked by human intronic and exonic sequences [117]. When
ubiquitously expressed using the actin-Gal4 driver, they did not detect antisense RNA foci but they observed an average of 42 sense nuclear RNA foci in glutamatergic neurons and 49 in glial cells. However, adult brain transcriptome analysis revealed that transcription was quite normal, strongly suggesting that sense RNA foci were not sufficient to alter global mRNA expression.
Depending on C9orf72 splice variant, the (G4C2) repeat is present in the promoter
region or within the first intron (Figure3). To answer whether the genomic location of the repeat could influence its toxicity, Moens et al. used two types of “RNA-only” constructs containing 100 copies of the repeat. One type of construct contained the sense or antisense repeats upstream of a polyA (sense/antisense polyA) and the other one contained the repeat within an artificial intron introduced into the GFP coding sequence (sense/antisense intronic) [118]. When expressed in adult Drosophila neurons, sense and antisense repeat-polyA constructs induced RNA foci within the cytoplasm, while RNA foci were mostly nuclear with both the sense and antisense intronic constructs. Despite the presence of these RNA foci, no effect on survival or climbing ability was observed indicating that they were not the primary cause of neurotoxicity [118]. However, this study did not investigate whether the co-expression of sense and antisense repeats could cause any detrimental effect. 3.3. Dipeptide Repeats Protein Toxicity
Despite its intronic localization, the (G4C2) expansion undergoes repeat-associated
non-AUG (RAN) translation from both sense and antisense transcripts to produce five dipeptide proteins (DPRs) [109,119,120]. The sense transcript gives rise to poly-glycine/ alanine (poly-GA) and poly-glycine/arginine (poly-GR) while the antisense transcript produces poly-proline/arginine (poly-PR) and poly-proline/alanine (poly-PA). Both sense and antisense transcripts produce poly-glycine/proline (poly-GP). To address whether these DPRs were toxic, Mizielinska et al. generated flies carrying transgene encoding “protein-only” by modifying the (G4C2) expansion with alternative codons [121]. Upon eye
expression, they showed that (PR)36and (GR)36were toxic but not (GA)36or (PA)36,
indi-cating that only arginine-rich DPRs were toxic. When 100-copy DPRs were expressed, the toxic effect of arginine-rich DPRs was exacerbated but the other DPRs did not show toxicity.
The toxicity of arginine-rich DPRs was also demonstrated by several other studies. For example, Drosophila expressing poly(PR)50in glutamatergic neurons, including
mo-toneurons, developed normally but were unable to escape out of the pupal case due to the absence of movement [122]. In contrast, flies expressing poly(GA)50 or poly(PA)50
developed normally [122]. The expression of an ATG-driven GFP-tagged poly(GR)50
in-duced eye degeneration that was not observed with ATG-driven GFP-tagged poly(GA)50
or poly(GP)47constructs [123,124]. At larval NMJs, the expression of poly(GR)100but not
poly(GA)100induced a decrease of presynaptic area, as well as a reduction in active zone
number [125].
Several DPRs are found in ALS patients; however, it remains to be determined whether some DPRs can interact with others to modulate toxicity. Indeed, one study has shown that (GR)80had a diffuse cytoplasmic localization in Drosophila salivary gland cells, while
(GA)80formed cytoplasmic inclusions [126]. Interestingly, when both (GR)80and (GA)80
were co-expressed, part of (GR)80formed cytoplasmic inclusions, suggesting that (GA)80
recruited (GR)80into these inclusions. Of note, the same observation was found in induced
pluripotent stem cell (iPSC)-derived human neurons [126]. The expression of (GR)80
induced cell loss at the wing margin that was partially suppressed by the co-expression of (GA)80. However, this protective effect of (GA)80was not observed in the eye [126]. Thus,
it is still unclear whether DPRs may interact with each other and modulate the toxicity of arginine-rich DPRs.
The transcellular spreading of misfolded proteins is thought to participate in the clinical progression of several neurodegenerative diseases. To answer whether DPRs have the ability to spread, mCherry-tagged DPRs were expressed in olfactory receptor neurons (ORNs) that target their axonal projections in fly brain. When 36 copies of (GA), (GR), or (PR) were used, no spreading was observed. In contrast, (GA)100 but not (GR)100or
(PR)100was detected in fly brains after 3 days of expression in ORNs [127]. Accordingly,
only (GA)100was detected in axons and synaptic terminals of ORNs. Interestingly, the
spreading was increased when a longer construct was used. It would be interesting to combine several DPRs to determine whether poly-GA may recruit other DPRs allowing them to spread between neurons.
In a way to understand how arginine-rich DPRs may cause toxicity, Lee et al. an-alyzed the interactome of GFP-tagged DPRs [124]. Among the identified interactors, 81 proteins were common to GFP-(PR)50and GFP-(GR)50and included some ALS-related
RBPs, such as TDP-43 and FUS, for example, suggesting that arginine-rich DPRs may alter RNA metabolism. Moreover, these common interactors showed enrichment in proteins containing low-complexity sequence domains (LCDs), which mediate the assembly of membrane-less organelles. By using an RNAi genetic screen targeting the orthologous genes in Drosophila, the authors found that most of the genetic modifiers of GFP-(PR)50
-induced toxicity were components of membraneless organelles such as nucleoli, the nuclear pore complex, and stress granules [124]. This strongly suggested that arginine-rich DPRs disturb the function of these organelles. In agreement with this, nucleocytoplasmic trans-port defects were also retrans-ported by other studies. A LOF genetic screen based on the modification of (G4C2)58-induced eye degeneration identified 18 modifiers that play a role
in nucleocytoplasmic transport, suggesting that nuclear retention of mRNA may be a cause of toxicity [123]. By performing a targeted RNAi genetic screen to identify modifiers of the rough eye phenotype induced by the expression of (PR)25, Boeynaems et al. showed
that nucleocytoplasmic transport was implicated in the pathogenic mechanism of C9orf72 HRE [128]. Another DPR interactome study was performed on flies expressing DPRs in adult brain neurons [129]. This study identified mostly ribosomal proteins as interactors of arginine-rich DPRs. A genetic screen based on the overexpression of ribosomal proteins or translation initiation factors revealed that the eukaryotic translation initiation factor 1A (eIF1A) mitigated the lifespan defects induced by (G4C2)36repeat or (GR)100pan-neuronal
expression. As eIF1A plays an important role in translation initiation, these data suggested that translation machinery might be affected by arginine-rich DPRs.
A recent study showed that poly(GR)80 expression in Drosophila muscles induced
alter-ations were the consequence of the entry of poly(GR)80within mitochondria where they
interacted with components of the mitochondrial contact site and cristae organizing system, leading to mitochondrial defects. Thus, it appears that poly(GR) may cause toxicity by interfering with mitochondrial functions, at least in muscle cells.
Altogether these studies have identified several pathways of C9orf72-associated toxi-city that may highlight novel therapeutic targets. For example, reducing the function of components of the specific machinery required for transcription of expanded repeat RNA, such as dDSIF and dPAF1 complexes, should reduce levels of toxic expanded RNA, as well as the production of toxic DPRs [110,111]. Another strategy would be to target the G-quadruplex structures of the expanded repeat RNAs. Indeed, the destabilization of the G-quadruplex structure by TMPyP4 treatments seems to be efficient as it suppressed HRE-induced neurodegeneration in the eye [115]. Furthermore, Simone et al. identified small molecules that specifically stabilize the G-quadruplex structures [131]. When administrated in the food, these small molecules decreased DPR production and improved survival of Drosophila expressing (G4C2)36repeats [131], confirming the therapeutic potential of this
approach. Lastly, the mitochondrial alterations induced by poly(GR) expression in muscles was rescued by treatment with nigericin, a K+/H+antiporter that rebalances mitochondrial matrix ion levels, opening the way to future potential therapeutic strategies [130].
4. FUS, an RNA-Binding Protein Associated with ALS
Fused in sarcoma, also known as translocated in liposarcoma (TLS), is a DNA/RNA-binding protein ubiquitously expressed. ALS-causative mutations in this RBP were discov-ered in 2009 [29,30]. In the nervous system, FUS is predominantly located in the nucleus and able to shuttle between the nucleus and the cytoplasm [132]. In patients, FUS is nuclear but FUS mutant forms are also found aggregated in the cytoplasm of neurons [29,30]. In ALS, the formation of protein aggregates is one hallmark of pathogenic mechanisms leading to motoneuron death, and the role of FUS in this process is important [65,133–135].
As an RBP, FUS has a pivotal role in many aspects of RNA metabolism and process-ing, including RNA splicing [136,137], transcription [138,139], nucleocytoplasmic trans-port [140], and translation (for a review, see [141]). FUS is a 526 amino acid protein encoded by 15 exons. The protein contains seven domains: an N-terminal region rich in glutamine, glycine, serine, and tyrosine residues (QGSY-rich domain), a prion-like domain, an RNA recognition motif (RRM), a nuclear export sequence (NES), two regions rich in arginine and glycine (RGG1 and RGG2), a zinc finger (ZnF) domain, and a C-terminal part containing a nonclassical PY-nuclear localization signal (NLS) motif [142] (Figure3). ALS-related mutations are located preferentially in the C-terminal part affecting the NLS domain of the protein. The pathogenic FUS mutations are missense changes, and the R521H and R521C mutations are the most commonly found in patients. As a result, in patient postmortem studies, a strong labeling of FUS was observed in neuron and glial cell nuclei but also as aggregates in the cytoplasm. Only few point mutations have been identified in the sequence encoding the N-terminal part or prion-like domain of the FUS protein [29,30,143,144]. More recently, mutations in the 30 untranslated transcribed region (UTR) of FUS have been described. Interestingly, these mutations affect the expression level of FUS rather than the cellular localization of the protein [145,146].
The mechanisms underlying the cause of neurodegeneration in FUS-ALS patients are still unknown even if numerous hypotheses have been made. Hence, mislocalization of FUS mutant forms in the cytoplasm leads to a disruption of FUS nuclear functions, such as tran-scription regulation or mRNA processing, and produces a LOF phenotype [134,147–151]. Conversely, the presence of FUS mutant forms in cytoplasmic aggregates creates a new toxic function of FUS in this compartment [144,152–154]. In an attempt to test these two hypotheses (LOF or new toxic GOF) to better understand the mode of action via which FUS mediates neurotoxicity, animal models for FUS-ALS have been generated. In vertebrate models, axonal degeneration and neuromuscular damages with protein aggregation in motoneuron were described as main phenotypes [148,155–157].
4.1. Drosophila Models of FUS-Related Neurodegeneration
In Drosophila, the first studies showed that eye-overexpression of wildtype FUS or different ALS-related mutated forms led to progressive neurodegeneration of the pho-toreceptors [147,158,159]. When expressed in the eyes, the phenotypes induced by the expression of wildtype FUS were a mild rough eye surface and a reduction in red pigment. The expression of different FUS mutants (R521C, 521H, 518K, R524S, or P525L) generated a more severe rough eye phenotype and even total depigmentation of the eyes (Table3).
Table 3.Gain-of-function phenotypes induced by FUS. CNS, central nervous system.
Gal4 Line UAS Line Phenotype/Reference
Act5C-Gal4 (ubiquitous)
FUS WT Lethal, no eclosion [160]
FUS R521G Lethal, no eclosion [160,161]
FUS R521H Lethal, no offspring [161]
FUS P525L Lethal, no offspring [161]
FUS∆32 No effect on viability [160]
Tubulin-Gal4 (ubiquitous)
FUS WT Lethal, no offspring [161]
FUS R521G Lethal, no offspring [161]
FUS R521H Lethal, no offspring [161]
FUS P525L Lethal, no offspring [161]
Tubulin-Gal4 Tubulin-Gal80TS (expression induced at
adult stage)
FUS WT Severe reduction of lifespan [161]
FUS R521G Severe reduction of lifespan [161]
FUS R521H Severe reduction of lifespan [161]
FUS P525L Severe reduction of lifespan [161]
Appl-Gal4 (pan-neuronal)
FUS WT Normal eclosion [158]
FUS R518K Pupal lethality [158]
FUS R521C Pupal lethality [158]
FUS R521H Pupal lethality [158]
Elav-Gal4 (pan-neuronal)
FUS WT
Rescued eclosion and locomotion in caz1mutants [159]
Lethal pupal [162]
Reduced viability at 25◦C, improved at 19◦C [160]
Lethal, no offspring [161]
FUS R521G Reduced viability at 25
◦C, improved at 19◦C [160]
Lethal, no offspring [161]
FUS R521H Lethal, no offspring [161]
FUS P525L Rescued eclosion in caz1mutants [159]
Lethal, no offspring [161]
FUS∆32 No effect on viability [160]
ElavGS
(pan-neuronal, inducible)
FUS WT RU486 treatment at eclosion, decline in lifespan, 50% lethality at 28 days [158]
Decreased lifespan, no degeneration in brain [162]
FUS R521C RU486 treatment at eclosion, 50% lethality at 10 days, impairment in climbing [158]
D42-Gal4 (motoneuron)
FUS WT
Reduced viability at 25◦C, improved at 19◦C, defect in adult climbing [160]
Mitochondria defect [163]
Defect in adult eclosion [164]
Eclosion defect and escapers with immature phenotype and reduced lifespan [161]
FUS R495X No effect on adult eclosion [164]
FUS R521G Reduced viability at 25
◦C, improved at 19◦C, defect in adult climbing [160]
Eclosion defect and escapers with immature phenotype and reduced lifespan [161]
FUS R521H Eclosion defect and escapers with immature phenotype and reduced lifespan [161]
FUS P525L Mitochondria defect [163]
Eclosion defect and escapers with immature phenotype and reduced lifespan [161]
Table 3. Cont.
Gal4 Line UAS Line Phenotype/Reference
D42-Gal4 Tubulin-Gal80TS (expression induced at
adult stage)
FUS WT Reduced lifespan and impaired flight ability [161]
FUS R521G Reduced lifespan and impaired flight ability [161]
FUS R521H Reduced lifespan and impaired flight ability [161]
FUS P525L Reduced lifespan and impaired flight ability [161]
OK371-Gal4 (glutamatergic neurons)
FUS WT
Mild larval crawling defect, no effect on synaptic boutons [158]
Disruption in motoneuron cluster, decreased synaptic bouton number, reduction in
mobility [147]
Larval locomotion activity impaired, decreased synaptic bouton number, pupal
lethality [160]
Reduced larval crawling, no effect on larval CNS size, 70% of adult eclosion [165]
Larval locomotion impaired, pupal lethality [164]
FUS R495X Normal larval locomotion, no defect in adult eclosion, no defect in climbing activity,
no effect on adult viability [164]
FUS R518K
Impaired larval locomotion, no effect on synaptic boutons, pupal lethality [158]
Drastic larval crawling reduction, reduction in larval CNS size, no adult
eclosion [165]
FUS R521C
Impaired larval locomotion, no effect on synaptic boutons, pupal lethality [158]
Drastic larval crawling reduction, reduction in larval CNS size, no adult
eclosion [165]
FUS R521G Larval locomotion activity impaired, decreased synaptic bouton number, pupal
lethality [160]
FUS R521H Impaired larval locomotion, no effect on synaptic boutons, pupal lethality [158]
FUS R524S Disruption in MNs cluster, decreased synaptic bouton number, reduction in mobility,
tail lifted [147]
FUS P525L
Disruption in MNs cluster, decreased synaptic bouton number, reduction in mobility,
tail lifted [147]
Larval locomotion impaired, pupal lethality [164]
FUS 4F-L (RRM Mutant)
No effect on larval crawling, no effect on the larval CNS size, no effect on adult
eclosion [165]
FUS∆32 No effect on larval locomotion, no effect on synaptic bouton number [160]
OK6-Gal4 (motoneurons)
FUS WT Increased synaptic bouton number [159]
Adult eclosion defect with immature escapers [166]
FUS R521G Adult eclosion defect with immature escapers [166]
FUS R521H Adult eclosion defect with immature escapers [166]
FUS R522G No effect on synaptic bouton number [159]
FUS P525L No effect on synaptic bouton number [159]
GMR-Gal4 (eye)
FUS WT
Very mild rough eye [158,165]
Reduction in red pigment, rough surface [147]
Malformed interommatidial bristles [162]
Severe rough eye phenotype [160]
Rough eye with pigment loss [164]
Reduced and rough eye [167]
FUS R495X Mild rough eye with slight pigmentation defect [164]
FUS R518K Rough eye [158,165,167]
FUS R521C Rough eye [158,160,165,167]
FUS R521G Rough eye [160]
FUS R521H Rough eye [158]
FUS R524S Severe rough eye [147]
FUS P525L Severe rough eye, depigmentation [147]
Rough eye with pigment loss [164]
FUS∆32Cter No phenotype [160]
FUS 4F-L (RRM Mutant)
Table 3. Cont.
Gal4 Line UAS Line Phenotype/Reference
CCAP-Gal4 (bursicon neurons)
FUS WT Adult eclosion impairment and escapers immature phenotype [166]
FUS R521G Adult eclosion impairment and escapers immature phenotype [166]
FUS R521H Adult eclosion impairment and escapers immature phenotype [166]
OK107-Gal4 (mushroom bodies)
FUS WT Thin mushroom body lobes [147]
FUS R524S Drastic decreased size of MB neurons, axonal degeneration [147]
FUS P525L Drastic decreased size of MB neurons, axonal degeneration [147]
MS1096-Gal4 (wing pouch)
FUS WT Defect in wing formation [160]
FUS R521G Defect in wing formation [160]
FUS∆32 No effect [160]
This table describes the phenotypes associated with different UAS FUS lines. Note that the experiments could be done at different temperatures and the UAS lines were generated using different genetic strategies (site-specific or random insertion) and tags.
As ALS is characterized by the degeneration of motoneurons, the action of FUS expres-sion in this specific neural population was considered. When expressed in motoneurons, all forms of FUS (wildtype and ALS-related) led to a deficit in locomotion at the larval stage followed by a lethality occurring at late pupal stage ([147,158] and Table3). At the NMJ level, the synaptic endings were altered. No consensus has emerged concerning a clear morphological change in the bouton numbers following FUS expression in either wildtype or mutants [147,158–160,168,169]. However, the synaptic boutons exhibited less and aberrantly organized active zones [168,169]. The post-synaptic compartment of the NMJs, like the clustering of the glutamate receptors, was altered when FUS was expressed in motoneurons [168]. As a result, the synaptic transmission was severely impaired at the NMJs. Electrophysiological studies showed that the amplitude of the excitatory junctional potentials was decreased in FUS mutant expression conditions, strengthening the idea that synaptic defects appear earlier than MN degeneration [168,169].
Adult eclosion defect and late pupal lethality were also observed using a general neuronal driver as Elav-Gal4. Nevertheless, use of inducible Elav-Gal4-GS line has allowed showing that FUS mutant forms induced a drastic climbing decline associated with a reduced lifespan, more severe than wildtype FUS did [158]. Thus, the FUS fly models recapitulate several characteristics found in ALS patients. Nevertheless, depending on the studies, some discrepancies can be noted concerning the severity of different phenotypes. It appears that the expression level of the different forms of FUS is a critical element that must be taken into consideration for the analysis and comparison of the observed phenotypes. Thus, the viability and the adult eclosion rate when FUS is expressed in motoneurons can vary, as well as the severity of the induced rough eye phenotype [160,168].
4.2. From FUS Endogenous Functions to Toxicity
4.2.1. Nuclear and Cytoplasmic Localization of a Shuttle Protein
The cellular localization of FUS has also been studied. As in ALS patients, mutated forms of FUS were both found in the nucleus and mislocalized in the cytoplasm, whereas wildtype FUS was always detected in the nucleus [29,30]. ALS-related FUS mutations are predominantly found in the C-terminal part of the protein containing the NLS do-main. These mutations led to a redistribution of the FUS protein in the cytoplasm in all system models used. Nuclear import of FUS is mediated by the nuclear transport recep-tor transportin (also known as karyopherin-β2) and impairment of this interaction led to FUS mislocalization in the cytoplasm [144,152,164]. This observation gave rise to the hypothesis that loss of physiological function of FUS in the nucleus contributes to the ALS pathology [170–172].
To decipher FUS functions, studies have focused on the role of the cabeza (caz) gene. caz is the only ortholog of FUS in Drosophila [173]. The caz1mutants appeared morphologically normal but displayed an adult eclosion defect, and the resulting adult escapers showed reduced lifespan and deficit in locomotion ([159] and Table4). Overexpression of FUS in
the nervous system of caz1Drosophila rescued the adult eclosion defect and restored both
lifespan and locomotor deficits. By contrast, mutated forms of FUS (P525L and R522G) acted on the survival to adulthood of caz1Drosophila with no effect on the lifespan and locomotion of adults [159].
Table 4.Phenotypes observed in cabeza mutants.
Mutant Line Phenotype Reference
caz1
No effect on NMJ morphology Adult eclosion impairment
Adult locomotion affected Reduced lifespan
[159]
Loss of ommatidia in the eyes [174]
caz2 Developmental delay and pupal lethality [175]
cazKO Developmental delay and pupal lethality [175]
cazlox, elav-Gal4, UAS Cre
(pan-neuronal)
Reduced adult offspring Adult motor deficit
Reduced lifespan
[175]
cazlox, Mef2-Gal4, UAS Cre
(muscles) Reduced muscle width [175]
CazFRT, elav-Gal4, UAS FLP
(pan-neuronal)
Reduced adult offspring
Reduced lifespan [175]
CazFRT, Mef2-Gal4, UAS FLP
(muscles) No effect [175]
This table describes the phenotypes associated with different cabeza mutant lines. Note that the experiments could be done at different temperatures.
The use of RNAi lines to knockdown caz gene expression in neurons showed that caz silencing did not affect lifespan but altered climbing performances ([150] and Table5). caz-knockdown in the eye induced a rough phenotype due to apoptotic cells in the pupal retina. Indeed, this phenotype could be rescued by the antiapoptotic p35 expression [176]. The en-gineering of new caz null mutant and conditional alleles using homologous recombination confirmed the previous phenotypes of pupal lethality and locomotor defects [175]. These results corroborated the fundamental role of caz in neural development and argue in favor of the LOF hypothesis to explain the FUS-induced neurodegeneration in ALS. caz GOF produced the same phenotypes as FUS overexpression (Table5). The larval locomotion was impaired with a reduced number of synaptic boutons and severe eye degeneration [160]. In addition, both Caz and FUS GOF induced apoptosis when expressed in motoneurons [150].
Both Caz and FUS are found in the nucleus in the nervous system. To address the role of FUS localization in toxicity, FUS constructs deleted for the NES domain were en-gineered in both wildtype FUS and ALS-related mutants. While FUSWT, FUS∆NES, and FUS∆NES R518Klocalized in the nucleus, FUSR518Kwas also found in the cytoplasm [158]. The double-mutant FUS∆NES R518Kbecame nontoxic, suggesting that FUS mutant cytoplas-mic localization is important for the toxicity to arise [158]. However, wildtype FUS deleted for its last 32 amino acids containing NLS domain (FUS∆32) localized outside of the nucleus but did not exhibit any phenotype when expressed in the eye or in motoneurons. Moreover, addition of an NLS sequence in the FUS∆32construct led to a nuclear localization of the protein associated with locomotor defect and eye degeneration phenotypes [160]. These contradictory results do not allow deciphering clearly the location (nucleus or cytoplasm, or both) where FUS induces its toxicity.
Table 5.Cabeza gain of function and RNAi-induced cabeza silencing phenotypes.
Gal4 Line Line Phenotype/Reference
Act5C-Gal4 (ubiquitous)
UAS-caz Lethal, no adult eclosion [160]
UAS-RNAi caz (363–399) VDRC 100291
Low adult eclosion [160]
No effect at 28◦C [150]
UAS-RNAi caz (1–167)
Lethal at 28◦C, no effect at 25◦C [150]
Late pupal lethality [176]
UAS-RNAi caz
(180–346) No effect at 28
◦C [150]
Elav-Gal4 (pan neuronal)
UAS-caz Almost lethal at pupal stage, few escapers [160]
UAS-RNAi caz (363–399) VDRC 100291
Adult eclosion defect [160]
No effect on lifespan, reduced mobility from young adult, decreased in
total axonal branch length [150] and synaptic bouton number [177]
Adult climbing defect, number of synaptic bouton and total branch
length reduced [178,179]
UAS-RNAi caz (1–167)
No effect on lifespan, reduced mobility from young adult. In larvae,
decreased synaptic bouton number [150,177] and decreased total axonal
branch length [150]
UAS-RNAi caz
(180–346) No effect on lifespan [150]
D42-Gal4 (motoneurons)
UAS-Caz Pupal lethality, no adult eclosion [160]
UAS-RNAi caz (363–399) VDRC 100291
Low adult eclosion [160]
OK6-Gal4
(motoneurons) UAS-caz
Increased synaptic bouton number [159]
Rescued the caz1phenotypes [159]
OK371-Gal4
(glutamatergic neurons) UAS-caz Severe larval crawling defect, reduction of synaptic bouton number [160]
GMR-Gal4 (eye)
UAS-caz Rough eye [160,177]
UAS-RNAi caz
(1–167) Rough eye [176,178]
UAS-RNAi caz (363–399) VDRC 100291
Severe rough eye [179,180] apoptosis [176]
nsyb-Gal4 (pan neuronal)
UAS-RNAi caz (363–399)
VDRC 100291 No significant lethality, slight motor defect [175]
UAS-RNAi HMS00790 No significant lethality, slight motor defect [175]
UAS-RNAi HMS00156 No significant lethality, slight motor defect [175]
Ppk-Gal4 (da neuron)
UAS-caz Reduced synaptic projections [181]
UAS-caz P398L Reduced synaptic projections [181]
This table describes the phenotypes associated with different UAS-cabeza or UAS RNAi cabeza lines. Note that the experiments could be done at different temperatures and the UAS lines were generated using different genetic strategies (site-specific or random insertion).
Interestingly, the observation that the overexpression of wildtype FUS or mutant forms in the motoneurons led to a drastic downregulation of cabeza points out that FUS could autoregulate its own expression in Drosophila [164,168]. Moreover, mutations affecting the 30-UTR region cause FUS overexpression and lead to ALS pathology [146], meaning that disruption of FUS autoregulation conducts to overexpression of wildtype FUS, which is sufficient to cause ALS.
These results strengthen (i) the notion of conservation between caz and FUS and (ii) the idea that the expression level of FUS is critical to trigger the neurodegeneration process [159,164,168]. Thus, elucidating the physiological functions of FUS with its different partners is necessary to better understand the mechanisms involved in the pathology.
4.2.2. FUS Alters Mitochondrial Physiology
Mitochondria continually undergo fission and fusion processes; the breakdown of this balance is decisive in neurodegenerative diseases. Mitochondria disruption has been exten-sively reported from ALS patient studies [182,183] (for a review, see [184]). In motoneurons, expression of FUSWTor ALS mutant FUSP525Lled to mitochondrial damages in transgenic flies. Compared to control condition, mitochondria were smaller, and the number of larger ones decreased. With FUS mutant overexpression conditions, the phenotype was more pronounced [163,185]. In addition, the mobility and the mitochondrial transport were reduced. Both anterograde and retrograde transports were affected; the frequency and du-ration of the transport interruption were increased, while the motile phase of mitochondrial transport was reduced [186]. Similar results were found using another neural model, the class IV dendritic arborization neurons (da neurons). Expression of FUS and Caz wildtype or mutant in da neurons altered the dendritic branching. FUSP525Lor CazP398Lmutants localized in the cytoplasm and were found at the synaptic projections of the da neurons, which were altered. All forms of FUS and Caz (wildtype or mutant) impaired axonal transport of the synaptic vesicles. A decrease in the number of synaptic mitochondria in da neurons was also observed. The expression of all forms of FUS and Caz induced an increase in the frequency of calcium transients in da neurons [181]. FUS was also found associated with mitochondria, and it interacts directly with HSP60, an ATPase dependent mitochondrial chaperone [187], which is involved in the translocation of FUS to the mito-chondria. Knocking down HSP60 expression via RNAi expression in motoneurons was sufficient to rescue the mitochondrial phenotypes induced by FUS overexpression [163]. The physiological role of FUS in mitochondria is still unknown but it seems that, in excess of FUS, the interaction of FUS with HSP60 promotes mitochondrial damage and toxicity. 4.2.3. FUS in the Nucleus Is Associated with Nuclear Bodies
The nucleus is compartmentalized in membraneless intranuclear compartments col-lectively named nuclear bodies (NBs). NBs include Cajal bodies, nucleoli, nuclear speckles, and paraspeckles that are dynamic structures responding to stress and controlling gene expression. NBs are the location of RNA biogenesis and maturation, and they are involved in the assembling of ribonucleoprotein complexes or in the retention of proteins [188–191]. NBs are composed of various proteins and RNAs, including the architectural RNAs (arcR-NAs) that are long noncoding RNAs (lncR(arcR-NAs) used as scaffolds [191–193]. Interestingly, FUS was found associated with paraspeckles, and both LOF and GOF of FUS caused dis-ruption of NBs [194]. In Drosophila, arcRNAs were found associated with hnRNPs. Notably, arcRNA hsrω is crucial for the formation of specific ω-speckles NBs. It also regulates the intranuclear trafficking and availability of different hnRNPs [195–199]. The loss of hsrω in neurons, using an RNAi-specific line, resulted in phenotypes closely related to the ones of caz LOF. Adults exhibited a shortened lifespan accompanied by locomotor deficit and a reduction in the number of synaptic boutons at the NMJs [179]. In addition, the subcellular localization of Caz changed and became cytoplasmic, leading to the conclu-sion that ω-speckles are involved in Caz compartmentalization. caz and hsrω genetically interact as the overexpression of hsrω enhanced the rough eye phenotype induced by caz expression [179]. The same interaction occurred between hsrω and FUS. Expressed in eye, hsrω RNAi rescued the toxicity induced by FUS. In this genetic condition, FUS was cytoplasmic as observed in the control condition, but punctate forms were detectable in the cytoplasm [179]. In this hsrω-knockdown background, FUS insoluble aggregates were not toxic and were found associated with Lysosome-associated membrane protein 1 (LAMP1), a marker of lysosomes that was upregulated [179,200]. Thus, misregulation of lncRNA could rescue the FUS-induced toxicity via the formation of nontoxic FUS aggregates through a mechanism that remains to be explored.
4.2.4. FUS Is a Multidomain Protein: Structure and Function
As mentioned above, FUS is composed of different domains whose functions have been revisited in recent years. Previously, the C-terminal domain was the focus of most studies because most ALS mutations cluster in the NLS sequence and were known to disrupt the nuclear localization of FUS. Revisiting the FUS amino-acid sequence showed that the N-terminal part contains a prion-like domain followed by a glycine-rich and an arginine–glycine–glycine repeat sequence (RGG) (Figure3). These three domains, which are composed of few different amino acids, are considered an LCD sequence. It was shown that the LCD of FUS is necessary for the formation of phase-separated liquid droplets or hydrogel [201–204]. To better understand the role of each FUS domain in the neurodegeneration process in vivo, systematic deletions or mutations of these domains have been generated in transgenic flies.
In motoneurons, expression of wildtype FUS led to pupal death. Flies cannot emerge from the pupal case and the few escapers observed display a soft cuticle and unexpanded wings that are characteristics of an immature phenotype (Table3). These phenotypes have been used as readout to determine the toxicity of the different domains of FUS. Bogaert et al. expressed in motoneurons different forms of FUS, deleted for distinct domains [161]. They analyzed the phenotypes obtained, comparing with the toxicity induced by wildtype FUS. In this screen, deletion of the Gly-rich domain, RGG1 domain, or zinc finger domain did not change the pupal death phenotype observed with wildtype FUS. This was also the case for the RRM domain, contrary to previous study reports ([165] and Table3). The conclusion was that the alteration of all these domains by themselves is not sufficient to drive FUS toxicity.
As already reported, mutations in the PY-NLS domain lead to mislocalization of FUS in the cytoplasm and induce a strong eye degeneration phenotype compared to wildtype FUS (Table3). In motoneurons, expression of FUS lacking its NLS domain partially allowed flies to eclose, contrary to wildtype FUS that induced pharate lethality. These emerging flies stayed immature, indicating that FUS lacking its NLS domain still confers some toxicity [161]. Unexpectedly, FUS lacking its NLS had a cytoplasmic localization but did not induced significant phenotype when expressed in the eye. However, the co-expression of this mutant with wildtype FUS enhanced the rough eye phenotype. In these flies, wildtype FUS was localized in the cytoplasm of retinal cells as aggregates, leading to the idea that expression of the C-terminally truncated FUS could interact with wildtype FUS to form toxic aggregates in the cytoplasm [205]. Moreover, PY-NLS was recently shown to have a function in disaggregation via the nuclear import receptor (NIR) Kapβ2. Indeed, NIRs act as chaperones to mediate nuclear import but also have an additional function in disaggregation activity and preventing fibrillization of RBPs in in vitro experiments [206]. Kapβ2 and FUS genetically interact, as Kapβ2-knockdown in the eye enhanced the rough eye phenotype induced by FUS overexpression [206].
An in vitro study showed that the LCD of FUS is involved in the self-assembly pro-cess [203]. In motoneurons, deletion of the QSGY domain (the most N-terminal part of the LCD) led to a partial rescue of the phenotypes induced by wildtype FUS expression. The adult eclosion rate of these flies was recovered, but the emerging adults showed an immature phenotype like the FUS-expressing escapers [161]. When expressed in the eye, an LCD-mutated form of FUS displayed no neurodegeneration phenotype and abolished the phenotype generated by mutations located in the NLS domain. Indeed, the self-assembly of FUS in the cytoplasm, via the LCD, seems indispensable to induce the neurodegeneration process in ALS [205].
The lack of phenotype observed with the C-terminally truncated FUS mutant could be explained by some differences in the LCD of Caz and FUS proteins. Indeed, the QGSY domain is absent in Caz protein [173]. Thus, NLS-mutated Caz did not display any phenotype when expressed in motoneurons, but addition of a QGSY domain to this NLS-mutant form generated a toxic protein and severe eclosion reduction, as observed for
